Skip to content

Conversation

@GuoningHuang
Copy link

@GuoningHuang GuoningHuang commented Jun 15, 2025

Add next-silu.mlir impement by tosa.sigmoid and tosa.mul as baseline;
Add manually optimized Silu operator next-silu-silu.mlir impement by vectorization optimization;
Integrated the optimization into the compilation pipeline as a pass by replacing the originally inefficient "sigmoid + multiplication" combination in linalg.generic with a vectorized SiLU implementation.

Copy link
Member

@zhanghb97 zhanghb97 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Giving quick feedback.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants